During 2002-2003, the main foci of the lab have been the improvement of MRI resolution, and the investigation of BOLD fMRI contrast mechanisms. In order to achieve improved resolution, new designs for multi-channel MRI technology have been developed and evaluated. It was determined that at 3.0 Tesla field strength, MRI performance in brain increases substantially with increasing channel number, until about 16 channels. Beyond this number, the incremental gains diminish and at this time do not outweigh the difficulties related to the increased system complexity. These findings were communicated to researchers of the major MRI equipment manufacturers. This optimal channel number is potentially larger for MRI systems at higher field strength, such as the recently installed 7.0 Tesla system. We are currently tranferring our multi-channel technology to this system to quantitatively evaluate perfomance differences. For the current 3.0 T 16-channel system, we have improved its performance for BOLD fMRI experiments by developing novel signal processing methods and acquisition strategies. For this purpose, a dedicated LINUX reconstruction cluster was built and parallel processing software was developed in-house. Using these novel technologies, we determined the optimal resolution for fMRI. It was found that the optimal resolution is around 1.5 mm at 3.0 Tesla. This improved resolution was used to allow detection of activation in subcortical brain structures such as the amygdala. Furthermore, high resolution fMRI studies were performed to investigate the temporal signal characteristics of the BOLD signal. Using m-sequence stimulation methodology, the timing of neurovascular control was investigated. It was found that the BOLD fMRI contrast mechanism adds several seconds of dispersion (blurring) to impulse neuronal activity. The amount of BOLD dispersion is reduced in rats. The fastest BOLD impulse responses were found in intracortical tissue in rats, putting an upper limit of 1.5 s to the temporal width (at half maximum) of the neurovascular control mechanism. In addition, it was determined that the BOLD contrast mechanism is to a large extent linear, which greatly facilitates fMRI experimental design and analysis. At a timescale of 1-10s, BOLD second order non-linearity is below 10% in human calcarine cortex. At 100-1000ms timescales, this non-linearity increases to about 20%. We have also started to compare measures of macroscopic electrical activity obtained with MEG with the BOLD signals. Our goal is to correlate BOLD fMRI with MEG signals to improve understanding of human brain function. The rationale is that sensory input and perception create distinctly different electrical signals, which both correlate with the BOLD fMRI signal. Preliminary experiments without visual input (obtained with subjects at rest with eyes closed) show fine-scale patterns of BOLD activity in human visual areas. These pattern appear to follow the functional boundaries as derived from sensory experiments. We are trying develop an understanding of this phenomenon by using carefully designed stimulation protocols. Furthermore, we are trying to determine how the patterns of electrical activity develop under these conditions. To this end, we have purchased an EEG system that will be installed in the MRI scanner in the coming year.